Treatment and Prevention of Heparin-Induced Thrombocytopenia

Transcription

1 CHEST Supplement ANTITHROMBOTIC THERAPY AND PREVENTION OF THROMBOSIS, 9TH ED: ACCP GUIDELINES Treatment and Prevention of Heparin-Induced Thrombocytopenia Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines Lori-Ann Linkins, MD ; Antonio L. Dans, MD ; COL Lisa K. Moores, MC, USA, FCCP ; Robert Bona, MD ; Bruce L. Davidson, MD, MPH, FCCP ; Sam Schulman, MD, PhD ; and Mark Crowther, MD Background: Heparin-induced thrombocytopenia (HIT) is an antibody-mediated adverse drug reaction that can lead to devastating thromboembolic complications, including pulmonary embolism, ischemic limb necrosis necessitating limb amputation, acute myocardial infarction, and stroke. Methods: The methods of this guideline follow the Methodology for the Development of Antithrombotic Therapy and Prevention of Thrombosis Guidelines: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines in this supplement. Results: Among the key recommendations for this article are the following: For patients receiving heparin in whom clinicians consider the risk of HIT to be. 1%, we suggest that platelet count monitoring be performed every 2 or 3 days from day 4 to day 14 (or until heparin is stopped, whichever occurs first) (Grade 2C). For patients receiving heparin in whom clinicians consider the risk of HIT to be, 1%, we suggest that platelet counts not be monitored (Grade 2C). In patients with HIT with thrombosis (HITT) or isolated HIT who have normal renal function, we suggest the use of argatroban or lepirudin or danaparoid over other nonheparin anticoagulants (Grade 2C). In patients with HITT and renal insufficiency, we suggest the use of argatroban over other nonheparin anticoagulants (Grade 2C). In patients with acute HIT or subacute HIT who require urgent cardiac surgery, we suggest the use of bivalirudin over other nonheparin anticoagulants or heparin plus antiplatelet agents (Grade 2C). Conclusions: Further studies evaluating the role of fondaparinux and the new oral anticoagulants in the treatment of HIT are needed. CHEST 2012; 141(2)(Suppl):e495S e530s Abbreviations: ACT 5 activated clotting time; aptt 5 activated partial thromboplastin time; CPB 5 cardiopulmonary bypass; CVA 5 cerebrovascular accident; DTI 5 direct thrombin inhibitor; ECT 5 ecarin clotting time; ELISA 5 enzymelinked immunosorbent assay; FDA 5 US Food and Drug Administration; GP 5 glycoprotein; GTI 5 Genetics Testing Institute; HIPA 5 heparin-induced platelet activation; HIT 5 heparin-induced thrombocytopenia; HITT 5 heparininduced thrombocytopenia with thrombosis; INR 5 international normalization ratio; LMWH 5 low-molecular-weight heparin; OD 5 optical density; PCI 5 percutaneous coronary intervention; PE 5 pulmonary embolism; PF4 5 platelet factor 4; RCT 5 randomized controlled trial; RR 5 relative risk; SC 5 subcutaneous; SRA 5 serotonin release assay; UFH 5 unfractionated heparin; VKA 5 vitamin K antagonist Summary of Recommendations Note on Shaded Text: Throughout this guideline, shading is used within the summary of recommendations sections to indicate recommendations that are newly added or have been changed since the publication of Antithrombotic and Thrombolytic Therapy: American College of Chest Physicians Evidence- Based Clinical Practice Guidelines (8th Edition). Recommendations that remain unchanged are not shaded For patients receiving heparin in whom clinicians consider the risk of HIT to be. 1%, we suggest that platelet count monitoring be performed every 2 or 3 days from day 4 to day 14 CHEST / 141 / 2 / FEBRUARY, 2012 SUPPLEMENT e495s

2 (or until heparin is stopped, whichever occurs first) (Grade 2C) For patients receiving heparin in whom clinicians consider the risk of HIT to be, 1%, we suggest that platelet counts not be monitored (Grade 2C) In patients with HITT, we recommend the use of nonheparin anticoagulants, in particular lepirudin, argatroban, and danaparoid, over the further use of heparin or LMWH or initiation/continuation of a vitamin K antagonist (VKA) (Grade 1C) In patients with HITT who have normal renal function, we suggest the use of argatroban or lepirudin or danaparoid over other nonheparin anticoagulants (Grade 2C). Remarks: Other factors not covered by our analysis, such as drug availability, cost, and ability to monitor the anticoagulant effect, may influence the choice of agent In patients with HITT and renal insufficiency, we suggest the use of argatroban over other nonheparin anticoagulants (Grade 2C) In patients with HIT and severe thrombocytopenia, we suggest giving platelet transfusions only if bleeding or during the performance Revision accepted August 31, Affiliations: From the Department of Medicine (Drs Linkins, Schulman, and Crowther), McMaster University, Hamilton, ON, Canada; the College of Medicine (Dr Dans), University of the Philippines Manila, Manila, Philippines; The Uniformed Services (Dr Moores), University of Health Sciences, Bethesda, MD; School of Medicine (Dr Bona), Quinnipiac University, North Haven, CT; and the University of Washington School of Medicine (Dr Davidson), Seattle, WA. Funding/Support: The Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines received support from the National Heart, Lung, and Blood Institute [R13 HL104758] and Bayer Schering Pharma AG. Support in the form of educational grants was also provided by Bristol-Myers Squibb; Pfizer, Inc; Canyon Pharmaceuticals; and sanofi-aventis US. Disclaimer: American College of Chest Physician guidelines are intended for general information only, are not medical advice, and do not replace professional medical care and physician advice, which always should be sought for any medical condition. The complete disclaimer for this guideline can be accessed at Correspondence to: Lori-Ann Linkins, MD, Department of Medicine, McMaster University, Juravinski Hospital, Rm-M0118, 1280 Main St W, Hamilton, ON, L8S 4K1, Canada ; linkinla@ mcmaster.ca. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( site/misc/reprints.xhtml ). DOI: /chest of an invasive procedure with a high risk of bleeding (Grade 2C) In patients with strongly suspected or confirmed HIT, we recommend against starting VKA until platelets have substantially recovered (ie, usually to at least /L) over starting VKA at a lower platelet count and that the VKA be initially given in low doses (maximum, 5 mg of warfarin or 6 mg phenprocoumon) over using higher doses (Grade 1C) We further suggest that if a VKA has already been started when a patient is diagnosed with HIT, vitamin K should be administered (Grade 2C). Remarks: We place a high value on the prevention of venous limb gangrene and a low value on the cost of the additional days of the parental nonheparin anticoagulant In patients with confirmed HIT, we recommend that that the VKA be overlapped with a nonheparin anticoagulant for a minimum of 5 days and until the INR is within the target range over shorter periods of overlap and that the INR be rechecked after the anticoagulant effect of the nonheparin anticoagulant has resolved (Grade 1C) In patients with isolated HIT (HIT without thrombosis), we recommend the use of lepirudin or argatroban or dana paroid over the further use of heparin or LMWH or initiation/continuation of a VKA (Grade 1C) In patients with isolated HIT (HIT without thrombosis) who have normal renal function, we suggest the use of argatroban or lepirudin or danaparoid over other nonheparin anticoagulants (Grade 2C). Remarks: Other factors such as drug availability, cost, and ability to monitor the anticoagulant effect may influence the choice of agent. The dosing considerations are the same as for patients with HITT (see section 3.2). For a recommendation on choice of nonheparin anticoagulants in the setting of renal insufficiency, see Recommendation In patients with acute HIT (thrombocytopenic, HIT antibody positive) or subacute HIT (platelets recovered, but still HIT antibody positive) who require urgent cardiac surgery, we suggest the use of bivalirudin over other nonheparin anticoagulants and over heparin plus antiplatelet agents (Grade 2C). e496s Treatment of HIT

3 In patients with acute HIT who require nonurgent cardiac surgery, we recommend delaying the surgery (if possible) until HIT has resolved and HIT antibodies are negative (see section 6.1) (Grade 2C). Remarks: Other factors not covered by our analysis, such as drug availability, cost, and ability to monitor the anticoagulant effect may influence the choice of agent. For recommendations for patients with a past history of HIT (. 3 months previous) who require cardiac surgery, see section In patients with acute HIT or subacute HIT who require percutaneous coronary interventions, we suggest the use of bivalirudin (Grade 2B) or argatroban (Grade 2C) over other nonheparin anticoagulants. Remarks: Other factors, such as drug availability, cost, and ability to monitor the anticoagulant effect, may influence the choice of agent In patients with acute or subacute HIT who require renal replacement therapy, we suggest the use of argatroban or dana paroid over other nonheparin anticoagulants (Grade 2C). Remarks: We acknowledge that the cost of argatroban may be prohibitive at some clinical centers. We further suggest that if the prothrombotic state of HIT appears to have resolved (as seen by normalization of the platelet count), saline flushes during dialysis would be a reasonable option. This suggestion is based on the presumed pathogenesis of thrombosis in this condition and not on the results of clinical trials In patients with a past history of HIT who require ongoing renal replacement therapy or catheter locking, we suggest the use of regional citrate over the use of heparin or LMWH (Grade 2C) In pregnant patients with acute or subacute HIT, we suggest danaparoid over other nonheparin anticoagulants (Grade 2C). We suggest the use of lepirudin or fondaparinux only if danaparoid is not available (Grade 2C). Remarks: Other factors, such as drug availability, cost, and ability to monitor the anticoagulant effect, may influence the choice of agent In patients with a history of HIT in whom heparin antibodies have been shown to be absent who require cardiac surgery, we suggest the use of heparin (short-term use only) over nonheparin anticoagulants (Grade 2C) In patients with a history of HIT in whom heparin antibodies are still present who require cardiac surgery, we suggest the use of nonheparin anticoagulants (see Recommendation 5.1.1) over heparin or LMWH (Grade 2C) In patients with a history of HIT in whom heparin antibodies have been shown to be absent who require cardiac catheterization or percutaneous coronary interventions, the recommended treatment is the same as in Recommendation In patients with a past history of HIT who have acute thrombosis (not related to HIT) and normal renal function, we suggest the use of fondaparinux at full therapeutic doses until transition to a VKA can be achieved (Grade 2C). This article offers recommendations on diagnosis and management of heparin-induced thrombocytopenia (HIT). Table 1 describes the question definition (ie, population, intervention, comparator, and outcome) addressed by the recommendations. 1.0 Methods and Overview of HIT We adhered to the general approach to developing recommendations described in the methodology article of these guidelines. 1 We searched the PubMed English language literature from January 1976 to June 2010 using the following search terms: heparin-induced thrombocytopenia, clinical trial, cohort, randomized clinical trial, argatroban, lepirudin, hirudin, bivalirudin, fondaparinux, diagnosis, laboratory assay, clinical prediction rule, platelet count monitoring, coronary artery bypass, cardiac surgery, cardiopulmonary bypass (CPB), angioplasty, transluminal percutaneous coronary, treatment, venous limb gangrene, platelet transfusion, renal replacement therapy, hemodialysis, hemofiltration, pregnancy, re-exposure, and recurrence. The primary efficacy outcome measures of interest were new thrombosis, limb amputation, major bleeding, and death (due to thrombosis or bleeding). In the cohort studies with historical controls, outcome events were counted if they occurred after treatment with the nonheparin anticoagulant was initiated, and from the date heparin was discontinued in the control group. 1.1 Value and Preferences Based on the relevant literature and the value and preference rating exercise conducted by the Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines panel, 2 we infer that from the patient s perspective, a venous thromboembolic event (eg, pulmonary embolism [PE], proximal DVT) carries similar weight as a major bleeding event (eg, gastrointestinal bleeding event), and that a stroke carries 2.5 times the weight of a major bleeding event. 1.2 Overview of HIT Pathogenesis of HIT: HIT is an adverse immunemediated drug reaction that is associated with a high risk of venous CHEST / 141 / 2 / FEBRUARY, 2012 SUPPLEMENT e497s

4 Table 1 [Introduction] Treatment and Prevention of HIT: Question Definitions PICO Question Section Population Intervention(s) Comparator Outcome(s) Methodology 2.1 Patients receiving heparin or LMWH for 5 d 3.1 Patients with strongly suspected or confirmed HIT with thrombosis Platelet count monitoring combined with the 4Ts Score 2.0 Screening for HIT No platelet count monitoring False negatives (increased risk of thrombosis if not treated with nonheparin anticoagulants) 3.0 Management of HITT (HIT with thrombosis) Discontinue heparin with or without starting a VKA Treatment with nonheparin anticoagulants False positives (increased risk of bleeding if treated with nonheparin anticoagulants) True negatives (do not have HIT) True positives (do have HIT) Decision analysis Death (thrombosis, bleeding) Cohort studies with Limb amputation historical controls 3.2 Treatment with nonheparin Treatment with other New thrombosis (arterial, venous) RCT, cohort studies anticoagulants nonheparin anticoagulants Major bleeding 3.3 Platelet transfusions No platelet transfusions Case series 3.4 Starting VKA before platelet recovery 3.5 Discontinuing thrombin inhibitor after minimum of 5 d of overlap with a VKA 4.1 Patients with strongly suspected or confirmed HIT without thrombosis No VKA until after platelet recovery Discontinuing thrombin inhibitor after, 5 d of overlap with a VKA 4.0 Management of isolated HIT (HIT without thrombosis) Discontinue heparin with or without starting a VKA 4.2 Treatment with nonheparin anticoagulants 5.1 Patients who require urgent cardiac surgery Treatment with nonheparin anticoagulants Treatment with other nonheparin anticoagulants Major bleeding 5.0 Management of patients with acute or subacute HIT in special situations Treatment with nonheparin anticoagulants Treatment with other nonheparin anticoagulants Venous limb gangrene Case series Limb amputation New thrombosis (arterial, venous) Secondary analysis of cohort studies Death (thrombosis, bleeding) Cohort studies with Limb amputation historical controls New thrombosis (arterial, venous) Cohort studies Death (thrombosis, bleeding) Cohort studies Limb amputation New thrombosis (arterial, venous) 5.2 Patients who require urgent PCI Cohort studies 5.3 Patients who require renal replacement therapy Major bleeding Case series Procedural success 5.4 Pregnant patients Case series (Continued) e498s Treatment of HIT

5 Table 1 Continued PICO Question Population Intervention(s) Comparator Outcome(s) Methodology 6.0 Management of patients with a past history of HIT Section Heparin or LMWH Nonheparin anticoagulants Recurrence of HIT Case series Reemergence of HIT antibodies 6.1 Patients who require cardiac surgery 6.2 Patients who require PCI Cohort studies 6.3 Patients who require VTE Case reports prophylaxis or treatment HIT 5 heparin-induced thrombocytopenia; HITT 5 heparin-induced thrombocytopenia with thrombosis; LMWH 5 low-molecular-weight heparin; PCI 5 percutaneous coronary intervention; PICO 5 population, intervention, comparator, and outcome; VKA 5 vitamin K antagonist. and arterial thrombosis. 3-6 Heparin exposure leads to the formation of IgG antibodies that recognize multimolecular complexes of platelet factor 4 (PF4) and heparin that form on the surface of platelets.7,8 These complexes bind to the FcɣIIa (IgG) receptors of platelets, 9,10 resulting in platelet activation and release of procoagulant, platelet-derived microparticles. 11,12 The end result is marked generation of thrombin and the formation of venous and arterial thromboses that are the clinical hallmark of HIT. Risk factors for HIT include duration and type of heparin exposure, 13 patient population, severity of trauma, 17 and gender. 18 Differences in the stoichiometry of heparin/pf4 complexes are believed to explain the 10-fold higher likelihood of HIT in patients who receive unfractionated heparin (UFH) compared with patients who receive low-molecular weight heparin (LMWH) or fondaparinux. 19,20 Patients who undergo cardiac or orthopedic surgery and receive UFH have a higher risk of HIT (1%-5%) than medical or obstetric patients (0.1%-1%). 13,14,21-24 Women have approximately twice the risk of developing HIT as men. 18 Table presents the incidence of HIT in various patient populations Clinical Features: Thrombocytopenia (defined as a platelet count, /L) is the most common clinical manifestation of HIT and occurs in 85% to 90% of patients. 4 If this definition is broadened to include a proportional fall in the platelet count (eg, 30%-50% fall even if the nadir remains /L), this increases to 90% to 95% of HIT cases. 4,5,27 The characteristic onset of the platelet count fall in HIT is 5 to 10 days after initiation of heparin (first day of heparin 5 day 0), particularly when heparin is administered perioperatively (typical-onset HIT). 36 Rapid-onset HIT refers to an abrupt platelet count fall (within 24 h) that occurs in patients who already have circulating HIT antibodies because of recent exposure to heparin (usually within the past month, occasionally as long as 100 days earlier). 36,37 Occasionally, thrombocytopenia can occur as long as 3 weeks after cessation of heparin (delayed-onset HIT). 38 Although thrombocytopenia is the most common presenting feature of HIT, in up to 25% of patients with HIT the development of thrombosis precedes the development of thrombocytopenia. 3,5 The pattern of thrombocytopenia following cardiac surgery using heparin is worthy of special mention. Although approximately 50% of patients who undergo cardiac surgery will develop HIT antibodies, only 1% to 2% will develop clinical HIT (thrombocytopenia with or without thrombosis). 14 In general, the platelet count falls by approximately 38% immediately after CPB (and continues to decline for the first 1-2 postoperative days before rising in a continuous fashion to a level above the preoperative Table 2 [Overview of HIT] Incidence of HIT According to Patient Population and Type of Heparin Exposure Patient Population (Minimum of 4-d Exposure) Incidence of HIT, % Postoperative patients Heparin, prophylactic dose 3,4,14, Heparin, therapeutic dose Heparin, flushes a LMWH, prophylactic or therapeutic dose 14, Cardiac surgery patients 14,27,28, Medical Patients with cancer 24, Heparin, prophylactic or therapeutic dose LMWH, prophylactic or therapeutic dose 26, Intensive care patients Heparin, flushes 33, 0.1 Obstetrics patients 21,22,34,35, 0.1 See Table 1 legend for expansion of abbreviations. a Case reports only. CHEST / 141 / 2 / FEBRUARY, 2012 SUPPLEMENT e499s

6 count).39,40 The following two patterns of thrombocytopenia should alert clinicians to the possibility of HIT following cardiac surgery: a fall in platelet count that begins. 4 days postoperatively (day of surgery 5 day 0), and thrombocytopenia that persists for. 4 days after surgery.41 The most common complication of HIT is venous thrombosis; 17% to 55% of untreated patients who present with thrombocytopenia develop DVT and/or PE.6,14,42 Arterial thrombotic events, including limb artery thrombosis, thrombotic stroke, and myocardial infarction (MI), also occur, but less often (from 3%-10%).6,43 After cardiac surgery, the majority of HIT-related thrombotic events are arterial.44,45 Approximately 5% to 10% of patients with HIT die, usually as a result of thrombotic complications.6,42 Less common manifestations of HIT include venous limb gangrene (5%-10% of patients with HIT with DVT treated acutely with a vitamin K antagonist [VKA] [eg, warfarin]),46 necrotizing skin lesions at heparin injection sites,47,48 adrenal hemorrhagic necrosis (due to adrenal vein thrombosis), and acute systemic reactions within 30 min of an IV heparin bolus injection (eg, fever/chills, tachycardia, hypertension, dyspnea, cardiopulmonary arrest).48 HIT can be complicated by disseminated intravascular coagulation severe enough to deplete fibrinogen.38,49 Despite severe throm- bocytopenia (but with a nadir rarely, /L), petechiae or other signs of bleeding are rarely seen.3 HIT is recognized as a clinicopathologic syndrome because diagnosis is based on the combination of a compatible clinical picture and the presence of platelet-activating anti-pf4 antibodies.50 Clinical prediction rules to assist physicians with determining the probability that a patient has HIT have been developed,51-55 the best studied of which is the 4Ts score (Fig 1) Evidence is emerging that patients with a low 4Ts score have a very low probability of HIT (0%-3%).51,56 However, many patients (24%-61%) with a high 4Ts score prove not to have HIT.51,56 Clinical assessment plays an essential role in the diagnosis of HIT for two reasons: (1) there is commonly a delay before the results of laboratory testing for HIT are available, and management decisions must be made immediately (the rate of thrombosis prior to treatment is approximately 5% per day)60; and (2) isolated HIT antibodies are both frequent and not diagnostic of HIT Laboratory Diagnosis of HIT: A large number of laboratory assays are currently used to diagnose HIT. A recent survey of specialized coagulation laboratories in North America identified eight different assays and wide discrepancies in practice between centers using the same assay.61 The assays can be divided into Figure 1. 4Ts score. *Timing of clinical sequelae, such as thrombocytopenia, thrombosis, or skin lesions. **Two points if necrotizing heparin-induced skin lesions even if thrombocytopenia not present. (Modified with permission from Warkentin and Linkins.59) e500s Treatment of HIT

7 two major categories according to the end point they measure: (1) antigen assays that detect the presence of HIT antibodies, and (2) functional assays that detect evidence of platelet activation (by HIT antibodies) in the presence of heparin. 62 Only a small proportion of patients who form HIT antibodies (seroconversion) will develop thrombocytopenia, and a smaller proportion will develop HIT-associated thrombosis. Antigen assays, the most commonly used being enzyme-linked immunosorbent assays (ELISA) that test for antibodies that are reactive against PF4/heparin or PF4/polyvinyl sulfonate, are very sensitive for HIT because they detect seroconversion. 15,63 However, not all of the antibodies these assays detect are capable of causing clinical HIT; hence, the specificity of these assays is only moderate. In contrast, functional assays, such as the serotonin release assay (SRA) and heparininduced platelet activation (HIPA), are sensitive and specific for HIT because they only detect antibodies that are capable of activating platelets. 15 The washed platelet SRA and HIPA are generally accepted as the reference standard assays for HIT. However, they are only available at a few centers because they are technically difficult, require human platelets from known reactive donors, and, in the case of the SRA, require working with radiation. 64,65 Most clinical centers use commercially available ELISAs because they do not have these limitations. The primary drawback of the ELISAs is their potential to overdiagnose HIT by detecting antibodies that are not pathogenic. ELISAs that only detect IgG antibodies appear to have better specificity for HIT (IgM and IgA antibodies are unlikely to cause HIT). 15,66 In patients recovering from HIT, there can be a lag time of several weeks between full platelet recovery and disappearance of the HIT antibodies (subacute HIT), particularly when using the ELISA for serologic testing. These patients are still at risk for developing rapid-onset HIT on heparin re-exposure (unless the washed platelet SRA or HIPA is negative and the ELISA is only weakly positive or strongly positive because of non-plateletactivating IgM or IgA antibodies). A class of commercial antigen assays that are designed to have a faster turnaround time than the ELISA (approximately 15 min vs 3.5 h [or days, if batched]) have entered the market. One of these assays, the ID-PaGIA Heparin/PF4 antibody test (DiaMed), is a gel centrifugation assay that uses the binding of antibodies to antigen-coated (PF4/heparin) high-density, red polystyrene beads. 67 This method can be performed in any blood bank that utilizes a gel centrifugation system for red cell antibody screening. The operating characteristics for this assay are reviewed in section Commercial Antigen Assays Compared With Reference Standard Assays for Diagnosis of HIT: To determine the accuracy of the commercially available antigen assays for HIT, we searched the literature for studies that: (1) compared the operating characteristics of these assays with at least one of the reference standard assays (ie, SRA or HIPA), and (2) used blood samples collected prospectively from consecutive patients with suspected HIT. The three studies that met our criteria evaluated two antigen assays: GTI-PF4 (Genetics Testing Institute [GTI]) and ID-PaGIA Heparin/PF4 antibody test (DiaMed AG) 56,68,69 ( Table 3 ). Both of these assays detect all classes of immunoglobulin. The sensitivity of the GTI-PF4 assay was 100% (if negative, HIT ruled out), whereas the specificity (82%-85%) was lower. Thus, many patients with positive tests, particularly those with moderate or low pretest Table 3 [Overview of HIT] Comparison of Commercial Antigen Assays With Reference Standard Assays Study/Year Population Intervention Comparator Outcomes Comments, % Bakchoul et al 68 /2009 Warkentin et al 69 / consecutive surgical and medical patients with suspected HIT 417 consecutive patients with suspected HIT (excludes 18 patients with indeterminate SRA or insufficient sample for ELISA testing) 213 consecutive patients with suspected HIT GTI-PF4 polyanion ELISA (OD. 0.4 units) PaGIA (DiaMed) (positive/negative) GTI-PF4 polyanion ELISA (OD. 0.4 units) HIPA GTI-PF4 ELISA GTI-PF4 ELISA Clinical assessment: 4Ts TP 35 of 35 PPV 528 score (Greifswald TN 376 of 465 NPV 5100 modification) FP 89 of 124 Sens 5100 HIT positive 5 HIPA FN 0 of 376 Spec 581 positive and high PaGIA PaGIA or intermediate TP 33 of 35 PPV 537 4Ts score TN 408 of 465 NPV 599 FP 57 of 90 Sens 594 FN 2 of 410 Spec 588 SRA (positive. 50% GTI-PF4 ELISA GTI-PF4 ELISA release) TP 41 of 41 PPV 543 HIT positive 5 SRA pos TN 309 of 364 NPV 5100 FP 55 of 96 Sens 5100 FN 0 of 309 Spec 585 Pouplard et al 56 /2007 GTI-PF4 SRA (positive. 20% GTI-PF4 ELISA GTI-PF4 ELISA polyanion ELISA release) TP 22 of 22 PPV 539 (OD. 0.4 units) TN 156 of 191 NPV 5100 PaGIA (DiaMed) HIT positive 5 SRA FP 35 of 57 Sens 5100 (positive/negative) pos FN 0 of 156 Spec 582 Clinical assessment: PaGIA PaGIA 4Ts score TP 21 of 22 PPV 557 TN 175 of 191 NPV 599 FP 16 of 37 Sens 595 FN 1 of 176 Spec 592 ELISA 5 enzyme-linked immunosorbent assay; FN 5 false negative; FP 5 false positive; GTI 5 Genetics Testing Institute; HIPA 5 heparin-induced platelet activation; NPV 5 negative predictive value; PF4 5 platelet factor 4; PPV 5 positive predictive value; Sens 5 sensitivity; Spec 5 specificity; SRA 5 serotonin release assay; TN 5 true negative; TP 5 true positive. See Table 1 legend for expansion of other abbreviation. CHEST / 141 / 2 / FEBRUARY, 2012 SUPPLEMENT e501s

8 probability, will not have HIT. The sensitivity of the PaGIA is lower than the GTI-PF4 (94%-95%), and the specificity is higher (88%-92%) than the GTI-PF4. In summary, both of these antigen assays can exclude a diagnosis of HIT, but neither assay is ideal as a stand-alone test to confirm the diagnosis of HIT. A negative ELISA or PaGIA in a patient with a low pretest probability of HIT excludes the diagnosis of HIT. A positive ELISA or PaGIA in a patient with a low pretest probability of HIT should not be interpreted as diagnostic for HIT and requires confirmation with a functional assay Commercially Available ELISA Using Manufacturer s Optical Density Threshold Compared With an Elevated Optical Density Threshold: There is a correlation between the strength of the reaction with an ELISA (measured using optical density units [OD]) and the likelihood of clinical HIT. 70,71 Three studies have addressed the question (retrospectively) of whether raising the OD threshold that is used to define a positive result with an ELISA would improve the specificity of the assay ,72 All three used the GTI-PF4 assay, which detects all classes of immunoglobulin (positive threshold set at 0.40 OD) ( Table 4 ). Two of the studies showed that raising the OD threshold to 1.0 increased the likelihood of a positive SRA result (specificity increased from 85% to 95%) 69 and increased the likelihood of new thromboembolic events (24% of patients at a threshold of 0.40 OD had a new thrombotic event compared with 59% at a threshold of 1.0 OD). 70 The third study showed that increasing the threshold of the GTI-PF4 to 1.20 OD and combining it with an intermediate or high 4Ts score identified all of the same HIT-positive patients as the SRA alone. 72 In summary, it appears that the combination of a threshold. 1.0 OD with a high clinical suspicion for HIT (eg, intermediate or high 4Ts score) may have a similar accuracy for diagnosing HIT as the reference standard assay (SRA). However, this strategy requires validation in prospective studies. For laboratories using the GTI-PF4 ELISA, we suggest reporting the quantitative value of the test result, together with the threshold used to define a positive result, over reporting the result only as positive or negative. For clinicians ordering the GTI-PF4 ELISA to determine whether a patient has HIT, we suggest taking into consideration both the pretest probability of HIT and the quantitative level of the GTI-PF4 ELISA result. A GTI-ELISA result between 0.40 and 1.0 OD in a patient with a low or moderate pretest probability for HIT should, if possible, be confirmed with a functional assay. 2.0 Screening for HIT 2.1 Platelet Count Monitoring Combined With the 4Ts Score for Patients Receiving Heparin/LMWH Platelet count monitoring is warranted when the benefits of early diagnosis and treatment of HIT exceed the potential harms of frequent platelet count monitoring, including cost, unnecessary anxiety and additional testing, unnecessary withdrawal of heparin, and the use of nonheparin anticoagulants with a higher bleeding risk. No studies have directly addressed the issue of whether advantages of platelet monitoring outweigh the disadvantages in patients receiving UFH/LMWH. Three retrospective studies showed a low rate of compliance with platelet count monitoring recommendations (4%-42%), a low rate of testing Table 4 [Overview of HIT] Studies Comparing Different OD Threshold Levels for Commercial ELISAs Study/Year Study Samples Participants Outcome ELISA OD Warkentin et al 69 /2008 Lo et al 72 /2007 Zwicker et al 70 / patients identified as HIT positive by SRA 16 patients identified as HIT positive according to different definitions (of laboratory and clinical criteria) 63 patients identified as HIT positive by PF4/hep polyanion ELISA (OD. 0.40) with clinical criteria determined by laboratory b HITT (n 5 19) HIT (n 5 22) Classic HIT: SRA. 50% 1, GTI-PF4 ELISA , IgG ELISA , 4Ts high or intermediate Liberal HIT: (GTI-PF4 ELISA. 0.40) Modified conservative: GTI-PF4 ELISA Ts high or intermediate Frequency SRA positive 50% release (95% CI) PF4/hep polyanion ELISA (GTI) 0% (0-1.2) OD, 0.4 3% ( ) OD % ( ) OD % ( ) OD % ( ) OD. 2.0 Clinical Events PF4/hep polyanion 11 of 16 (69%) TECs (2 TEC postdiagnosis) 12 of 32 a (37%) TECs (2 TEC postdiagnosis) Identified same 16 patients as classic HIT definition ELISA (GTI) Median, 2.39 (IQR, ) Median, 0.89 (IQR, ) Median, 2.39 (IQR, ) Within 30 d c Clinical events PF4/hep polyanion ELISA (GTI) Thrombosis (n 523) OD. 1: 59% TECs Mean, 1.41; SD, 0.87 No thrombosis (n 540) OD, 1: 24% TECs (P 5.01) OD. 1: 36% TECs OD, 1: 9% TECs (P 5.07); OR, 5.7; 95% CI, Mean, 0.80; SD, 0.46 IQR 5 interquartile range; OD 5 optical density; TEC 5 thromboembolic complication. a Includes 16 patients identified by the classic HIT definition and an additional 16 patients (one of the additional patients had a TEC). b Recent platelet count, /L, platelet count of 50% in setting of heparin therapy or a prior history of HIT. c Includes the 15 patients initially diagnosed with HITT and eight patients with HIT who developed thrombosis within 30 d of diagnosis with HIT. e502s Treatment of HIT

9 for HIT antibodies in patients who became thrombocytopenic (5%-19%), and a low rate of initiation of a nonheparin anticoagulant when the suspicion of HIT was high enough to warrant laboratory testing (0%-55%) The findings of the above studies suggest previous recommendations for platelet monitoring have not been widely implemented. 76 Furthermore, when platelet count monitoring is done and platelets drop, heparin is not necessarily stopped nor is a nonheparin anticoagulant started. Possible reasons for these findings include the burden of platelet count monitoring, the limited availability of laboratory assays for serological confirmation of HIT, expense associated with using nonheparin anticoagulants, and a lack of awareness of the guidelines. We conducted a decision analysis to determine the reduction in HIT-related thrombotic events that could be achieved in an ideal setting if the recommendations for platelet count monitoring, laboratory testing for HIT, and initiation of a direct thrombin inhibitor (DTI) were all followed. To reduce the potential for expensive testing and inappropriate treatment of patients with a low clinical probability of HIT, we assumed that platelet count monitoring would be done as part of a clinical assessment of the patients probability of HIT using the 4Ts score (see Table 5 for key model assumptions; Table 6 for data sources and model inputs). Table 7 outlines the summary of findings for this decision analysis. This decision analysis shows that, in an ideal setting, for every 1,000 patients screened with platelet count monitoring and application of the 4Ts Score, the best estimate suggests one episode of thrombosis will be prevented, at the cost of one major bleeding event (although CIs for both thrombosis and bleeding overlap no effect) ( Table 7 ). Both the baseline risk of thrombosis (ie, patient population and type of heparin) (see Table 2 ) and the availability of the HIT assays Table 5 [Section 2.1.1] Key Model Assumptions for Platelet-Monitoring Decision Analysis Platelet count monitoring is intended to identify patients with isolated HIT; therefore, patients who initially present with HIT-related thrombosis were excluded. Patients with a moderate or high 4Ts score will have heparin discontinued, an ELISA ordered, and argatroban started, whereas patients with a low 4Ts score will be assumed not to have HIT and will continue to receive heparin. Patients with a positive ELISA will have an SRA performed for confirmation. Patients with a positive SRA will continue to receive treatment with argatroban, and patients with a negative SRA will resume prophylaxis with heparin (assuming 100% sensitivity and specificity of the SRA for HIT). The results of the HIT assays are available within 24 h of being ordered. See Table 1 and 3 legends for expansion of abbreviations. Table 6 [Section 2.1.1] Data Sources and Model Inputs for Platelet Monitoring Decision Analysis We selected a high-risk setting for HIT for this analysis: postoperative orthopedic surgery patients who are receiving UFH for thromboprophylaxis (incidence of HIT. 1%). 14 The likelihood ratios for low, moderate, and high 4Ts scores were derived from the study by Lo et al. 51 The sensitivity and specificity of the ELISA were assumed to be 100% and 86%, respectively. 15,52 The assumptions for rates of thrombosis are as follows: (A) Thrombosis rate in patients with HIT who received argatroban 5 argatroban arm of a pooled analysis of historical controlled studies (0.069). 76 (B) Thrombosis rate in patients with HIT who did not receive argatroban 5 control group of a pooled analysis of historical controlled studies (0.224). 76 (C) Thrombosis rate in patients without HIT who did not receive argatroban 51 of 37 of assumption (B) because HIT is reported to increase the risk of thrombosis by 37-fold (0.22/ ). 3 (D) Thrombosis rate in patients without HIT who were treated with argatroban 5 onethird of Assumption (C) as derived from the hazard ratio in patients treated with argatroban (0.002). 76 HIT was assumed not to influence the major bleeding rate independently of treatment with nonheparin anticoagulants. Consequently, the major bleeding rate in those treated with argatroban was the same as the major bleeding rate in patients with HITT treated with argatroban in the pooled argatroban studies (0.08), and the major bleeding rate in those who were untreated was the same as the major bleeding rate in the control arm of the pooled argatroban studies (0.022). 76 Sensitivity analyses were performed using different sensitivities and specificities for the ELISA and SRA, and different assay availability (ie, only ELISA available, no HIT assays available). UFH 5 unfractionated heparin. See Table 1 and 3 legends for expansion of other abbreviations. influence the benefit-to-risk ratio of platelet count monitoring. Clinical centers that do not have access to the reference standard assays will have a higher number of false-positive results and consequently a higher proportion of major bleeding events (ie, when the ELISA is the only HIT assay available, two episodes of thrombosis are prevented at the cost of major bleeding events for every 1,000 patients screened). Another factor that influences the feasibility of platelet monitoring with 4Ts screening is the cost. Although individual platelet counts are inexpensive, the cost of the HIT assays and nonheparin anticoagulants can be substantial (eg, in a formal costeffectiveness analysis the cost of treating one patient with HIT with argatroban for 5 days was estimated at $3,500-$4,500 in the United States 2004 prices). 77 The estimated cost in 2011, for the drug alone, for 5 days is $5,000 US. The issues with respect to platelet count monitoring outlined above were discussed at the American College of Chest Physicians meeting in February Criticisms of the decision analysis included the use of bleeding estimates based on doses of argatroban that are no longer used, delay in obtaining prompt results even at centers that have the SRA or HIPA available, and the potential for harm in missing cases of HIT CHEST / 141 / 2 / FEBRUARY, 2012 SUPPLEMENT e503s

10 Table 7 [Section 2.1.1] Summary of Findings for Platelet Count Monitoring Decision Analysis: Should Platelet Count Monitoring Be Performed in Patients Who Receive Heparin or LMWH for 5 d? Outcomes New thrombosis Major bleeding No. of Participants (Studies) Follow-up No studies available No studies available Quality of the Evidence (GRADE) Very low due to uncertainty of model assumptions Very low due to uncertainty of model assumptions Relative Effect RR, RR, RR, RR, Anticipated Absolute Effects, Time Frame for All Outcomes 30 d Risk Without Monitoring Risk Difference With Monitoring SRA and ELISA both available 8.2 thrombotic events fewer thrombotic per 1,000 events per 1,000 Only ELISA available 8.2 thrombotic events fewer thrombotic per 1,000 events per 1,000 SRA and ELISA both available 22 bleeding events more bleeding events per 1,000 per 1,000 Only ELISA available 22 bleeding events more bleeds per 1,000 per 1,000 The basis for the risks are provided in Tables 5 and 6. The anticipated absolute effect is expressed as risk difference, and is based on the baseline risk in the comparison group and the relative effect of the intervention. High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. GRADE 5 Grades of Recommendations, Assessment, Development, and Evaluation; RR 5 risk ratio. if platelet count monitoring is not performed. The attendees voted in favor, by a small margin, of providing specific recommendations regarding platelet count monitoring despite the uncertainty of the benefitto-risk ratio of this practice. It should be noted that for each of the recommendations below,. 20% of the attendees voted in the opposite direction. Recommendations For patients receiving heparin in whom clinicians consider the risk of HIT to be. 1% ( Table 2 ), we suggest that platelet count monitoring be performed every 2 or 3 days from day 4 to day 14 (or until heparin is stopped, whichever occurs first) (Grade 2C) For patients receiving heparin in whom clinicians consider the risk of HIT to be, 1% ( Table 2 ), we suggest that platelet counts not be monitored (Grade 2C). 2.2 Platelet Count Monitoring in Patients Recently Treated With Heparin/LMWH Platelet count monitoring for HIT in patients who have recently been exposed to heparin or LMWH differs from that described above because the timing of onset of HIT in these patients differs. If a patient still has circulating HIT antibodies from a previous exposure to heparin (typically within the past days), re-exposure can lead to a large platelet count fall within 24 h. 36 As with typical-onset HIT, there are no studies evaluating the benefit-to-risk ratio of this approach. Obtaining a baseline platelet e504s count prior to initiating anticoagulant therapy for a patient with VTE is considered standard medical practice; however, obtaining a platelet count 24 h later can be difficult because of the widespread use of outpatient LMWH therapy. Statement 2.2 : For patients exposed to heparin within the past 100 days, we suggest that a baseline platelet count be obtained prior to starting heparin or LMWH therapy, and that a repeat platelet count should be drawn 24 h later, if feasible. 2.3 Platelet Count Monitoring in Patients With Acute Inflammatory Reactions After IV Heparin Bolus Rarely, a patient who is given an IV heparin bolus will develop an acute inflammatory reaction (eg, fever, chills) and/or cardiorespiratory symptoms (eg, hypertension, tachycardia, dyspnea, chest pain, cardiorespiratory arrest) within 30 min of drug administration. These acute systemic reactions are strongly suggestive of acute HIT. 48 Statement 2.3: For patients who present with acute systemic reactions within 30 min of an IV heparin bolus, we suggest performing a platelet count. 3.0 Management of HIT Complicated by Thrombosis 3.1 Discontinue Heparin or Initiate VKA vs Treatment With Nonheparin Anticoagulants The first step in the treatment of HIT complicated by thrombosis (HITT) is discontinuation of all forms of heparin and LMWH (including heparin flushes and heparin-coated catheters). Whether taking this step alone is enough to prevent further thrombotic Treatment of HIT

11 Table 8 [Section 3.1] Summary of Findings for Argatroban for Treatment of HITT: Should Patients With HITT Receive Argatroban Over Discontinuing Heparin and/or Starting a VKA? Anticipated Absolute Effects, Time Frame 37 d for All Outcomes Outcomes No. of Participants (Studies) Follow-up Quality of the Evidence (GRADE) Relative Effect (95% CI) Risk With Discontinue Heparin/Start VKA Risk Difference With Argatroban (95% CI) Death due to 419 (2 cohorts) 37 d b Very low due to thrombosis a risk of bias and imprecision Limb amputation New thrombosis 419 (2 cohorts) 37 d b Very low due to risk of bias and imprecision 419 (2 cohorts) 37 d b Moderate due to risk of bias, but with large effect Major 419 (2 cohorts) 37 d b Very low due to bleeding c risk of bias and imprecision RR, 0.12 ( ) 152 deaths per 1, fewer deaths per 1,000 (from 100 fewer to 145 fewer) RR, 1.26 ( ) RR, 0.45 ( ) RR, 3.70 ( ) 109 amputations per 1, thrombotic events per 1, major bleeding events per 1, more amputations per 1,000 (from 51 fewer to 216 more) 191 fewer thrombotic events per 1,000 (from 101 fewer to 250 fewer) 59 more major bleeding events per 1,000 d (from 10 fewer to 554 more) The anticipated absolute effect is expressed as risk difference (and its 95% CI) and is based on the baseline risk in the comparison group and the relative effect of the intervention (and its 95% CI). For evidence profile, see Table S3. High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. PRBC 5 packed RBCs. See Table 1 and 7 legends for expansion of other abbreviations. a As judged by the investigators b Follow-up was 30 d past cessation of treatment in patients receiving argatroban and 37 d from baseline in control patients. c Defined as a hemoglobin drop of at least 20 g/l or requirement for 2 units of PRBC or an intracranial hemorrhage or a bleed into a joint. d There were three fatal bleeding events in patients who received argatroban (HIT and HITT combined). complications secondary to HITT has been evaluated in pooled analyses of prospective cohort studies with historical controls. 76,78 DTIs lepirudin and argatroban have each been compared with historical controls who received the best available care at the time ( Tables 8, 9 ). In the majority of cases, best available care consisted of discontinuation of heparin alone or substitution of heparin with a VKA. An overview of the methodology of these studies is available in Table S1 (tables that contain an S before the number denote supplementary tables not contained in the body of the article and available instead in an online data Table 9 [Section 3.1] Summary of Findings for Lepirudin for Treatment of HITT: Should Patients With HITT Receive Lepirudin Over Discontinuing Heparin and/or Starting a VKA? Anticipated Absolute Effects, Time Frame 35 d for All Outcomes Outcomes No. of Participants (Studies) Follow-up Quality of the Evidence (GRADE) Relative Effect (95% CI) Risk With Discontinue Heparin/Start VKA Risk Difference With Lepirudin (95% CI) Limb amputation New thrombosis 289 (3 cohorts) 35 d Very low due to risk of bias and imprecision 289 (3 cohorts) 35 d Moderate due to risk of bias, but with large effect Major 289 (3 cohorts) 35 d Very low due to risk of bleeding b bias and imprecision RR, 0.70 ( ) 80 amputations per 1, fewer amputations per 1,000 (from 58 fewer to 64 more) RR, 0.28 ( ) RR, 2.31 ( ) 253 thrombotic events per 1, major bleeding events per 1, fewer thrombotic events per 100 a (from 122 fewer to 215 fewer) 87 more major bleeding events per 1,000 c (from 4 fewer to 314 more) The anticipated absolute effect is expressed as risk difference (and its 95% CI) and is based on the baseline risk in the comparison group and the relative effect of the intervention (and its 95% CI). For evidence profile, see Table S4. High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. See Tables 1, 7, and 8 for expansion of abbreviations. a There were three deaths due to thrombosis in patients who received lepirudin (HITT and HIT combined). b Defined as a fatal bleeding event or an intracranial hemorrhage or a bleeding event that led to permanent disability or requirement for 2 units of PRBC. c There were five fatal bleeding events in patients who received lepirudin (HITT and HIT combined). CHEST / 141 / 2 / FEBRUARY, 2012 SUPPLEMENT e505s